Customer support chatbot

Widget & chat API · Vercel · Pro

• vcpu = app_vcpu = 0.5
• ram_gb = app_ram_gb = 0.5
• Pro list price $20.00/month
• note: includes $20 of usage credit; function, bandwidth and build usage beyond it is not modeled

AI framework · Vercel AI SDK

• free open-source software (Apache-2.0); runs inside the app

Chat model · OpenAI API · gpt-5-mini

• input_tokens = conversations_per_month * messages_per_conversation * (top_k * chunk_tokens + prompt_overhead_tokens) = 2,000 × 4 × (4 × 500 + 1,200) = 25,600,000
• output_tokens = conversations_per_month * messages_per_conversation * answer_tokens = 2,000 × 4 × 200 = 1,600,000
• input: 25,600,000 tokens ÷ 1M × $0.25 = $6.40
• output: 1,600,000 tokens ÷ 1M × $2.00 = $3.20

Embedding model · OpenAI API · text-embedding-3-small

• input_tokens = sync_runs_per_month * changed_articles_per_sync * tokens_per_article + conversations_per_month * messages_per_conversation * question_tokens = 30 × 5 × 900 + 2,000 × 4 × 40 = 455,000
• input: 455,000 tokens ÷ 1M × $0.02 = $0.0091

Knowledge index · Pinecone · Starter

• vector_gb = help_articles * tokens_per_article / chunk_tokens * vector_kb_per_chunk / 1000000 = 250 × 900 / 500 × 8 / 1000000 = 0.0036
• write_units = sync_runs_per_month * changed_articles_per_sync * tokens_per_article / chunk_tokens * vector_kb_per_chunk = 30 × 5 × 900 / 500 × 8 = 2,160
• read_units = conversations_per_month * messages_per_conversation * read_units_per_query = 2,000 × 4 × 1 = 8,000
• write_gb = sync_runs_per_month * changed_articles_per_sync * tokens_per_article / chunk_tokens * vector_kb_per_chunk / 1000000 = 30 × 5 × 900 / 500 × 8 / 1000000 = 0.00216
• query_tb = conversations_per_month * messages_per_conversation * help_articles * tokens_per_article / chunk_tokens * vector_kb_per_chunk / 1000000 / 1000 = 2,000 × 4 × 250 × 900 / 500 × 8 / 1000000 / 1000 = 0.0288
• returned_gb = conversations_per_month * messages_per_conversation * returned_kb_per_query / 1000000 = 2,000 × 4 × 20 / 1000000 = 0.16
• Starter list price $0.00/month

Knowledge sync · Trigger.dev · Hobby

• job_runs = sync_runs_per_month = 30
• job_compute_seconds = sync_runs_per_month * help_articles * sync_seconds_per_article = 30 × 250 × 0.5 = 3,750
• executions = sync_runs_per_month * (changed_articles_per_sync + 2) = 30 × (5 + 2) = 210
• Hobby list price $10.00/month
• compute seconds (Small 1x): 3,750 × $0.0000338 = $0.13
• runs: 30 ÷ 10,000 × $0.25 = $0.00075
• usage $0.13 − included credit $10.00 = $0.00

Conversation store · Neon · Launch

• db_gb = conversations_per_month * messages_per_conversation * retention_months * kb_per_message / 1000000 = 2,000 × 4 × 12 × 2 / 1000000 = 0.192
• cu_hours = db_cu * db_active_hours = 0.25 × 300 = 75
• compute CU-hours: 75 × $0.106 = $7.95
• storage GB-month: 0.192 × $0.35 = $0.07
• note: Launch has no base fee; the $15 on the page is a typical bill

Cache & rate limits · Upstash · Pay as You Go

• commands = conversations_per_month * messages_per_conversation * cache_commands_per_message = 2,000 × 4 × 3 = 24,000
• Pay as You Go: metered plan, no base fee
• commands: 24,000 ÷ 100,000 × $0.20 (catalog: Pay as You Go) = $0.05

Handoff email · Postmark · Basic

• emails = conversations_per_month * handoff_rate = 2,000 × 0.15 = 300
• Basic list price $15.00/month
• emails: 300 − 10,000 free = 0 ÷ 1,000 × $1.80 = $0.00

LLM observability · Langfuse · Core

• units = conversations_per_month * messages_per_conversation * observations_per_message = 2,000 × 4 × 3 = 24,000
• requests = conversations_per_month * messages_per_conversation * 2 = 2,000 × 4 × 2 = 16,000
• Core list price $29.00/month
• units (traces + observations + scores): 24,000 − 100,000 free = 0 ÷ 100,000 × $8.00 = $0.00